Research On Rate Control Of Scalable Video Coding

Scalable video coding as the extension of the H.264/AVC standard, can produce a bitstream which could support different frame rates, resolutions and video qualities through one coding process. This feature of scalable video coding can well meet the current transmission network and the diverse receiving devices. Rate control plays an important part in current video coding systems. Although rate control has not been strictly defined in any video compression standards, but related research has always been the most popular topic of video coding area. Since the rate control of scalable video coding has its own characteristics, the research on this subject has a certain practical significance.This paper first introduces the concept of the scalable video coding. Then the classical rate control of the past standards and the algorithm JVT-G012which was proposed in the H.264/AVC was described. After summarised the rate control strategy which was used in the current proposed scalable video coding reference software JSVM, finally we give the technologies used in this paper and the details of our algorithm.In this rate control algorithm, by using the switched MAD prediction model, not only can get more accurate predicted MAD when it changes abruptly due to high motion or scene changes, but also can take full advantage of the temporal and the spatial information. Considering the characteristics of the hierarchical B-frames prediction structure, in the process of the frame layer bit allocation, different weighting factors were set to different hierarchical levels. By given the different rate-distortion models to the different types of frames in different hierarchical levels(I frames and P frames using the quadratic rate-distortion model, B frames using the R-Qstep rate-distortion model), thus we can earn more accurate description of the relationship between rate and distortion.